Performence Tuning
Performence Tuning
Performence Tuning
Introduction
This document is a guideline process that, where possible, should be followed when reviewing SAP system performance and tuning. This document assumes knowledge of the BASIS module within SAP environments.
This process covers the process of reviewing system performance and where applicable identifying issues, problems or bottlenecks that can or will lead to poor system performance. This document can also be used to help identify the cause of poor system performance.
A SAP R/3 system is an online transactional processing (OLTP) system. Good system performance is essential to the business and users to get the most out of the system. Regular proactive checks on system performance will help to identify any potential issues and ensure good system performance is available to those users.
SM50 is also a snapshot view of current activities but specific to the instance you are logged onto. This shows all work processes configured for the instance.
If all DIA work processes are in constant use this will lead to high wait time and SAP will hang.
If a problem is detected, the data in the workload monitor can be used as follows to identify the area of the system where the problem is located. First check for general performance problems affecting all transactions. Good general performance is normally indicated by: Wait time < 10% response time Average roll-in time < 20ms Average roll wait time < 200ms Average load (and generation ) time < 10% of response time (< 50ms) Average database request time < 40% of (response time wait time) Average CPU time < 40% of (response time wait time) Average CPU time not much than processing time Average Dialog response time Depends on customer requirements there is no general rule but below 2 seconds is expected. Analysis Roadmap: Using the Workload Monitor Wait time > 10% of response time? General performance problem? High Database time > 40% of response time?
Detailed analysis of the database Processing time > CPU time x 2? Detailed analysis of hardware bottlenecks Load time > 50 ms? Detailed analysis of R/3 memory configuration (Is the program buffer too small?) Roll wait time or GUI time > 200 ms? Detailed analysis of interfaces and GUI communication.
From detailed analysis > last minutes load > transaction profile enables you to find out the most used transactions and the average response time for these transactions. Tuning these transactions creates the greatest improvement in overall performance.
Analysis Roadmap: Using the Transaction Profile Transaction profile sorted by Response time total Programs with high CPU time: CPU time >40% of response time
Programs with high database time: database time > 40% of response time
Detailed analysis of SQL statements with SQL Trace Programs with high GUI times (> 200ms) Network Check
Operating system configuration parameters CPU Bottlenecks are indicated during several snapshots by: Idle CPU <10% Load Average : N processes waiting in front of the CPU Note: You must refresh the screen several times, as each display is a momentary snapshot. Memory bottlenecks are indicated during several snapshots by: An increase in page outs for UNIX or page ins for NT
For the instance on which the user is logged on, ST02 shows the percentage usage of buffers and memory. Good System performance or good tuning is dictated by: Good buffer hit ratio > 97% Few buffer swaps < 5000 per day Some free space in buffer but not too much Reasons for program buffer swaps could be that the program buffer is sized too small. This will lead to programs are swapped out, and the reload causes high load and long database request times. The solution is to increase the program buffer size.
The lower part of ST02 displays data on extended memory size and usage. No heap memory should be in use, this will lead to performance problems as physical memory is exhausted and SAP will be paging into operating system swap. Memory utilization (Extended memory) Current use,<Maxuse<<In Mem
For those systems which use application servers and employee logon groups a large imbalance in the amount of users logged onto the application servers could lead to performance problems on that particular instance. Make sure that no one application server has proportionally many more users that any other. Low number of users on Central Instances is good.
6. Run the report in se38: RSTS0043 to clear the logs: spool, background job logs.
7. Do spool consistency check in SP12 ->TemSe Data Storage -> consistency check.